EDP Sciences logo
Subscriber Authentication Point
Search
Menu
Home All issues Volume 650 (June 2021) A&A, 650 (2021) A90 Full HTML
Free Access

Table A.2.

Comparison of photo-z estimates.

Reference Method (a) Bias(b) rms (c) Fraction of outlier in %
1 (trainZ) −0.2086 0.1808 0
ANNz2 0.00063 0.0270 4.4
BPZ −0.00175 0.0215 3.5
Delight −0.00185 0.0212 3.8
EAZY −0.00218 0.0225 3.4
FlexZBoost −0.00027 0.0154 2.0
GPz 0.00000 0.0197 5.2
Lephare −0.00161 0.0236 5.8
METAPhoR 0.00000 0.0264 3.7
CMNN −0.00132 0.0184 3.5
SkyNet −0.00167 0.0219 3.6
TPZ 0.00309 0.0161 3.3
2 Convolutional neural network(CNN) 0.0001 0.0456 (d) 0.31
3 METAPHOR −0.004 0.065 0.98
ANNz2 −0.008 0.078 1.60
BPZ −0.020 0.048 1.13
4 CNN + density field (mode) 0.0038 (d) 0.83
CNN + density field (median) 0.0045 (d) – 0.44
CNN + density field (mean) 0.0066 (d) 0.31
5 kNN −0.0001 ± 0.0 0.0165 ± 0.0001 4.0
6 kNN 0.001 (e) 0.36 10.7 (f)
7 DEmP (g) −0.0291 0.1018 0.16
DEmP (h) −0.0175 0.07 0.17
8 Trees and random forest(Regression mode) −0.00008 0.0225 0
Trees and random forest (Classification mode) 0.00218 0.0246 0
9 ArborZ −0.006 (e) 0.985 1.9
10 GP-GL 0.0946 0.1420 5.3
GP-VL 0.828 0.1251 5.5
GP-VC 0.0294 0.0435 4.7
11 Ensemble of ANNs, trees and KNN (nominal solution) 0.0002 0.034 0.105
Ensemble of ANNs, trees and KNN(⟨PDF⟩) 0.00035 0.034 0.105
Ensemble of ANNs, trees and KNN(PDF) 0.00035 0.052 0.1
12 PS1-STRM (All validation) base estimate 0.0003 0.0342 2.88 (i)
PS1-STRM (All validation) Monte-Carlo sampled 0.0010 0.0344 2.99
PS1-STRM (Non-extrapolated) base estimate 0.0005 0.0322 1.89
PS1-STRM (Non-extrapolated) Monte-Carlo sampled 0.0013 0.0323 2.00

Notes. Values are provided where information was available.

(a)

Acronyms are defined in the respective literature;

(b)

Bias: defined as mean of Δz = (zp − zs)/(1 + zs);

(c)

rms((zp − zs)/(1 + zs));

(d)

σMAD = 1.4826 × MAD, where MAD (Median Absolute Deviation) is the median of |Δz − Median(Δz)|;

(e)

Average of δz = zp − zs;

(f)

fraction of outliers defined as number of objects with |Δz| > rms(Δz)±0.5;

(g)

exclusively using wide-band photometry from Wide fields of HSC (https://hsc.mtk.nao.ac.jp/ssp/) as additional photometric input;

(h)

exclusively using deep photometry from Deep and UltraDeep fields of HSC as additional photometric input;

(i)

fraction of outliers defined as number of objects with |Δz| > 0.15.

References. (1) Schmidt et al. (2020); (2) Pasquet et al. (2019); (3) Amaro et al. (2019); (4) Shuntov et al. (2020); (5) Graham et al. (2018); (6) Curran (2020); (7) Nishizawa et al. (2020); (8) Carrasco Kind & Brunner (2013); (9) Gerdes et al. (2010); (10) Almosallam et al. (2016); (11) Sadeh et al. (2019); (12) Beck et al. (2021).

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.